Title
Modeling Future Biofuel Supply Chains using Spatially Explicit Infrastructure Optimization
Permalink https://escholarship.org/uc/item/5qw9j3xh Author Parker, Nathan Publication Date 2011
eScholarship.org Powered by the California Digital Library
Institute of Transportation Studies ◦ University of California, Davis One Shields Avenue ◦ Davis, California 95616
PHONE (530) 752-6548 ◦ FAX (530) 752-6572 www.its.ucdavis.edu
Research Report – UCD-ITS-RR-11-04
Modeling Future Biofuel Supply
Chains using Spatially Explicit
Infrastructure Optimization
January 2011
-i-
Modeling Future Biofuel Supply Chains using Spatially Explicit
Infrastructure Optimization
by
NATHAN CHANDLER PARKER B.S. (Wake Forest University) 2001 M.S. (University of California, Davis) 2007
DISSERTATION
Submitted in partial satisfaction of the requirements for the degree of DOCTOR OF PHILOSOPHY
in
Transportation Technology and Policy in the
OFFICE OF GRADUATE STUDIES of the
UNIVERSITY OF CALIFORNIA DAVIS
Approved:
__________________________ Joan M. Ogden, Chair __________________________ Bryan M. Jenkins __________________________ Yueyue Fan Committee in Charge 2011
-ii- Abstract
Policies have been enacted that promote biofuels with the goal of reducing greenhouse gas emissions, reduce dependence on petroleum and to spur rural economic growth. The supply of biofuels that can meet these three goals is limited. The cost of this supply is influenced by the geography of the biomass resource and demand for fuels. Existing studies projecting the future supply have not accounted for the spatial aspects of the biofuel supply in detail.
This dissertation presents a spatially-explicit model of future biofuel supply chains in the United States, with the goal of providing supply curves of biofuels by resource-technology pathway with detailed accounting of the required infrastructure. The model is used to analyze the potential supply of biofuels for meeting the federal Renewable Fuel Standard (RFS2) and
analyze biofuels from waste and residue resources in California at high resolution with accounting for air pollutant emissions.
The results of the national case study project that domestic biofuels can achieve the RFS2 mandates for 2022 at fuel prices of between $3.4 and $5 per gasoline gallon equivalent. The largest sources of variation are the cost of cellulosic biofuel technologies and the availability of low cost waste resources. Building the 200-250 cellulosic biorefineries needed to achieve the target requires a capital investment greater than $100 billion but less than $360 billion depending on technology development and choice of cellulosic technology.
-iii-
Waste and residue biomass can provide quantities of biofuels that assist with policy goals. Nationally, waste and residue resources are projected to provide between 35 and 64 percent of the RFS2 mandate in both 2018 and 2022. In California, biofuels from waste and residue resources have limited potential for petroleum displacement, but could contribute 40-70% of the LCFS emissions reductions with mixed and uncertain results on air quality.
-iv- Acknowledgements
This dissertation and my graduate studies were enlightened by the mentoring I received from my advisors and the lively discussions with my fellow graduate students. I would like to thank: Prof. Joan Ogden for her inquisitiveness and encouragement; Prof. Bryan Jenkins for his enthusiasm for this research and for assembling the team that made the national
modeling work possible; and Prof. Yueyue Fan for being my mentor in optimization modeling.
I would also like to thank my collaborators who made this work possible: Dr. Quinn Hart for his brilliance in dealing with the data I threw at him and making it easy on me; Peter Tittmann for always trying to find a better way; Richard Nelson for his advise and assistance with agricultural resiudes; Ed Gray, Ken Skog, and Pat Hu for providing data that made this research possible.
This research was funded by Chevron Technology Ventures, LLC, United States Department of Energy and the United States Department of
Agriculture. I am grateful for their suport. In addition, I would like to thank the sponsers of the Sustainable Transportation Energy Pathways (STEPS) Program at the Institute of Transportation Studies at the University of California, Davis (ITS-Davis) for support during my graduate studies.
Finally, I would like to thank my friends and family for keeping my life fun and interesting.
-v-
TABLE OF CONTENTS
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vi-
,":"#! =.1>0($9.1*/0=31.8.2)*=3'('=/0(9?'/9.1* ,+!
,":"!! (0$.%(=0*'$$0$$&01/$* :#!
,":",! <./01/9'8*A9.(07910()*8.='/9.1$* :!! ,":":! /('1$<.(/'/9.1*10/5.(6*&.408* ::! 93A! 1*'"0"0,%$7*%(>>.&>."($*%#>($"(!%.*#&!-$"&0%'&.%(0(!?#"#% ;A!
;! 2&05*.#"&0%$*270&!&,"*#% ;C! ;3/! 2&.0%*$7(0&!% AD! ;36! '($$?%(2"1%4*$7?!%*#$*.%E'(4*F%)"&1"*#*!%(01%G.*0*@()!*H%1"*#*!%I% 7?1.&$.*($4*0$%&'%!">"1#% A9! ;39! 2*!!-!&#"2%*$7(0&!% AJ! ;3;! '"#27*.K$.&>#27%1"*#*!% =9!
;3A! 2&4>(."#&0%&'%2*!!-!&#"2%$*270&!&,"*#% =A! ;3=! 2&05*.#"&0%$*270&!&,?%4&1*!%>(.(4*$*.#% =C! :";"#! 0D9$/912*7'=989/90$* ;C! A! 1($(% =B! A3/! .*#&-.2*%(##*##4*0$#% =B! @"#"#! '2(9=%8/%('8*(0$94%0$* BE! @"#"!! 7.(0$/*(0$94%0$* B+! @"#",! <%8<5..4* C!! @"#":! &%19=9<'8*$.894*5'$/0$* C:! @"#"@! 010(2)*=(.<$* CC! @"#";! =.&&.49/)*=(.<$* +:! @"#"B! '19&'8*7'/$*'14*5'$/0*2(0'$0$* +;! @"#"C! $%&&'()*.7*1'/9.1'8*(0$.%(=0*4'/'*$0/* +C!
-vii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
-viii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
-ix- C",",! 8916$*/.*'2(9=%8/%('8*0=.1.&9=*&.408* #++! C",":! %1=0(/'91/)* #++! J3;! '-$-.*%.*#*(.27%1".*2$"&0#% 6DD! .*'*.*02*#% 6D;! (>>*01"+%(L%'**1#$&2<%!&,"#$"2#%2&#$%4&1*!% 6/=! (>>*01"+%)L%(11"$"&0(!%*4"##"&0#%1*$("!#% 666! (>>*01"+%2L%4&1*!%2&1*% 66C!
-x-
LIST OF TABLES
TABLE 1: MAJOR BIOFUELS RELATED POLICIES... 2
TABLE 2: MODEL VARIABLES AND INDICES ... 28
TABLE 3: MODEL PARAMETERS ... 29
TABLE 4: SUMMARY OF CONVERSION TECHNOLOGY STATUS AND COST MODELS USED ... 47
TABLE 5: PARAMETER DEFINITION FOR LEVELIZED COST EQUATION ... 49
TABLE 6: STANDARDIZED ASSUMPTIONS FOR COMPARING TECHNOLOGIES ... 49
TABLE 7: ECONOMIC PARAMETERS FOR DRY MILL CORN ETHANOL TECHNOLOGY ... 53
TABLE 8: ECONOMIC PARAMETERS FOR LIPIDS TO DIESEL CONVERSION TECHNOLOGIES ... 58
TABLE 9: ECONOMIC PARAMETERS FOR CELLULOSIC ETHANOL CONVERSION TECHNOLOGIES... 62
TABLE 10: ECONOMIC PARAMETERS FOR FISCHER-TROPSCH DIESEL CONVERSION TECHNOLOGIES... 65
TABLE 11: MODEL PARAMETERIZATION FOR CONVERSION FACILITIES.... 68
TABLE 13: GRAIN AND RESIDUE YIELD GROWTH FROM 1940 TO 2000 ... 71
TABLE 14: NUTRIENT REPLACEMENT FOR AGRICULTURAL RESIDUES ... 77
TABLE 15: SUMMARY OF MSW LANDFILLED ... 85
TABLE 16: ROADSIDE COST OF GRAIN AND LIPID RESOURCES... 99
TABLE 17: SUMMARY OF PROJECTED BIOMASS RESOURCES IN 2018... 102
TABLE 18: SUMMARY OF OTHER RESOURCE ASSESSMENTS ... 104
TABLE 20: RFS2 MANDATED QUANTITY OF BIOFUELS ... 116
TABLE 21: SUMMARY OF THE PARAMETER CHANGES IN SCENARIOS ... 127
TABLE 22: SUMMARY OF BASELINE AND RESOURCE SENSITIVITY SCENARIOS FOR MEETING 2018 RFS2 MANDATE ... 146
-xi-
TABLE 23: SUMMARY OF TECHNOLOGY SCENARIOS ... 148
TABLE 24: CONVERSION EFFICIENCY OF CELLULOSIC ETHANOL TECHNOLOGY ... 157
TABLE 25: CAPITAL COST PARAMETERS FOR BIOREFINERIES... 164
TABLE 26: PROCESS DESCRIPTION... 165
TABLE 27: OPERATING COST PARAMETERS... 165
TABLE 28: ADDITIONAL ECONOMIC PARAMETERS... 166
TABLE 29: POTENTIAL TO EMIT FOR VARIOUS PROPOSED BIOREFINERIES ... 170
TABLE 30: EMISSION FACTORS FOR CONVERSION TECHNOLOGIES ... 172
TABLE 31: SUMMARY OF BIOFUEL PATHWAY PERFORMANCE... 184
TABLE 32: EMISSION FACTORS FOR FERTILIZERS... 222
TABLE 33: EMISSION FACTORS FOR MSW CLASSIFICATION ... 223
TABLE 34: EMISSION FACTORS FOR DIESEL CONSUMED BY MODE... 224
TABLE 35: POTENTIAL TO EMIT BY PROCESS UNIT FOR PROPOSED BLUEFIRE MSW-TO-ETHANOL BIOREFINERY ... 225
TABLE 36: POTENTIAL TO EMIT FOR PROPOSED VERENIUM ENERGYCANE-TO-ETHANOL BIOREFINERY ... 226
-xii-
LIST OF FIGURES
FIGURE 1: MODEL ORGANIZATION AND INTERACTION OF SUBMODELS ... 26
FIGURE 2: SCHEMATIC OF OPTIMIZATION MODEL... 27
FIGURE 3: IMPACT OF LINEARIZING THE CONVERSION COST MODELS ON AVERAGE AND MARGINAL COSTS ... 41
FIGURE 4: POTENTIAL LOCATIONS FOR BIOREFINERIES... 44
FIGURE 5: COMPARISON OF ESTIMATED LEVELIZED COSTS OF PRODUCTION FOR CORN ETHANOL ... 52
FIGURE 6: COMPARISON OF ESTIMATED LEVELIZED COST OF PRODUCTION FOR FAME BIODIESEL... 55
FIGURE 7: COMPARISON OF ESTIMATED LEVELIZED COST OF PRODUCTION FOR HYDROTREATMENT OF LIPIDS TO DIESEL... 57
FIGURE 8: COMPARISON OF ESTIMATED LEVELIZED COST OF PRODUCTION FOR CELLULOSIC ETHANOL ... 61
FIGURE 9: COMPARISON OF ESTIMATED LEVELIZED COST OF PRODUCTION FOR F-T DIESEL TECHNOLOGIES... 64
FIGURE 10: COMPARISON OF MODELED CELLULOSIC BIOFUEL TECHNOLOGIES USING SWITCHGRASS AS FEEDSTOCK ... 66
FIGURE 11: MAPS OF AGRICULTURAL RESIDUES IN THE HISTORICAL AND HIGH CASES... 79
FIGURE 12: MAP OF FOREST RESIDUE RESOURCES ... 82
FIGURE 13: MAP OF PULPWOOD SUPPLY... 83
FIGURE 14: DISTRIBUTION OF ESTIMATED MSW PRODUCTION ... 88
FIGURE 15: MAPS OF POTENTIAL ENERGY CROP SUPPLIES UNDER THE BASELINE AND HIGH ENERGY CROP SCENARIOS ... 94
FIGURE 16: LIPID RESOURCES INCLUDING CANOLA AND SOY CRUSHING PLANTS AND ANIMAL FATS FROM RENDERING FACILITIES... 98
FIGURE 17: BASELINE SUPPLY OF CELLULOSIC BIOMASS RESOURCES. ... 99
FIGURE 18: CELLULOSIC BIOMASS SUPPLY IN HIGH SCENARIO... 100
-xiii-
FIGURE 20: TRANSPORTATION COST AS A FUNCTION OF DISTANCE BY MODE... 107 FIGURE 21: NATIONAL TRANSPORTATION NETWORK. ... 110 FIGURE 22: DISTRIBUTION OF FUEL TERMINALS AND PROJECTED 2015
VMT BY CENSUS TRACT ... 112 FIGURE 23: MODELED SERVICE AREAS FOR EACH FUEL DISTRIBUTION
TERMINAL... 113 FIGURE 24: BASELINE BIOFUEL SUPPLY CURVE ... 130 FIGURE 25: FUEL PATHWAYS FOR MEETING 2018 RFS2 MANDATE UNDER
THE BASELINE ASSUMPTIONS ... 132 FIGURE 26: DISTRIBUTION OF F-T DIESEL BIOREFINERY SIZES IN
BASELINE... 134 FIGURE 27: BASELINE SUPPLY CURVE BY PATHWAY ... 136 FIGURE 28: SUPPLY CURVES FOR RESOURCE SENSITIVITY SCENARIOS . 139 FIGURE 29: BIOFUEL PATHWAYS FOR MEETING 2018 RFS2 MANDATE IN
HIGH AND LOW FEEDSTOCK SCENARIOS ... 140 FIGURE 30: SUPPLY CURVES FOR TECHNOLOGY SENSITIVITY SCENARIOS
... 141 FIGURE 31: SUPPLY CURVES FOR THE FUEL DEMAND SENSITIVITY
SCENARIOS WITH THE ETHANOL DOMINANT TECHNOLOGY SCENARIO ... 144 FIGURE 32: ACCUMULATION OF INPUTS AND EMISSIONS ALONG THE
SUPPLY CHAIN... 159 FIGURE 33: BIOMASS COST, VOLUME AND NUMBER OF DISCRETE SUPPLY
POINTS BY TYPE... 161 FIGURE 34: LOCATION OF BIOMASS RESOURCES - FOREST, AGRICULTURAL AND MUNICIPAL... 162 FIGURE 35: TRANSPORTATION NETWORK SHOWING THE CONNECTIONS
TO THE RESOURCE LOCATIONS AND THE SET OF POTENTIAL
LOCATIONS... 163 FIGURE 36: POTENTIAL SUPPLY OF BIOFUELS BY PATHWAY ... 173 FIGURE 37: BREAKDOWN OF LEVELIZED COST OF FUEL BY STAGE OF THE
-xiv-
FIGURE 38: REQUIRED CAPITAL INVESTMENT FOR A GIVEN ANNUAL
PRODUCTION OF BIOFUEL BY PATHWAY... 175 FIGURE 39: EMISSIONS BY PATHWAY NORMALIZED BY CALIFORNIA
REFORMULATED GASOLINE WITHOUT CREDIT FOR ELECTRICITY PRODUCTION ... 177 FIGURE 40: EMISSIONS BY PATHWAY NORMALIZED BY CALIFORNIA
REFORMULATED GASOLINE EMISSIONS WITH CREDIT FOR
ELECTRICITY PRODUCTION DISPLACING NATURAL GAS COMBINED CYCLE ELECTRICITY... 178 FIGURE 41: BREAKDOWN OF EMISSIONS OF CO, N20, NOX AND PM2.5... 179 FIGURE 42: CARBON INTENSITY OF BIOFUEL PATHWAYS ... 180 FIGURE 43: OPTIMAL CONFIGURATION FOR BIOFUEL PRODUCTION FOR
FOREST AND WOOD WASTES AT $3.25/GGE - F-T DIESEL AND ETHANOL ... 181 FIGURE 44: OPTIMAL CONFIGURATION FOR BIOFUEL PRODUCTION FROM
MSW - F-T DIESEL AND ETHANOL... 182 FIGURE 45: OPTIMAL CONFIGURATION FOR BIOFUEL PRODUCTION FROM
AGRICULTURAL STRAWS AND STOVERS - F-T DIESEL AND ETHANOL ... 182!
1
INTRODUCTION
1.1 MotivationProduction of biofuels is increasing worldwide and especially in the United States (OECD-FAO, 2010). This increase is driven by many factors but chief among them are policies mandating or promoting biofuels (see Table 1). These policies promote biofuels as a means to reduce petroleum dependence, reduce greenhouse gases and spur rural economic development. Looking to the future, the policies in place call for transformative change and growth in the biofuels sector. The dominant biofuel technology in the United States, corn ethanol, has little room to grow within the existing federal mandate, while a nascent cellulosic biofuel industry is required to grow from today’s demonstration units to commercial production of 16 billion gallons per year (BGY) in 12 years. Where this fuel will come from, at what cost and with what economic, environmental, and land use impacts are questions without satisfactory answers to date. This dissertation presents a spatially explicit model of future biofuel supply chains in the United States, with the goal of providing answers to these key policy questions.
Table 1: Major biofuels related policies Policy Jurisdiction Summary Renewable Fuels
Standard United States 36 billion gallons of renewable fuels mandated by 2022 in 4 categories Fuel Quality
Directive European Union 10% of energy used in transport must be from renewable sources by 2020 and a 6% reduction in GHG emissions in transport
Low Carbon Fuel
Standard California The average carbon intensity of fuels sold must be reduced by 10% by 2020.
Spatial features of energy supplies are generally simplified to national or regional averages in existing assessments future energy supplies. All could be improved with greater detail to the spatial layout of the energy
infrastructure with biomass the spatial features are especially important as it has low yield (Btus/acre) compared to other energy sources and costly transport due to low energy density. Furthermore, the generally low
economic value per mass of energy feedstock biomass relative to agricultural commodities leads to greater importance of the transportation costs than is generally considered in models of agricultural production (Searcy et al., 2007). For these reasons, existing tools for considering either agricultural production or energy supply are likely to be inadequate for a realistic analysis of biofuel supply.
1.2 Research questions
The focus of this dissertation is the development of a modeling
methodology explicitly incorporating the spatial aspects of the biofuel supply chain. Using a spatially explicit framework, I seek to estimate the cost,
quantity, direct land use and lifecycle greenhouse gas emissions of biofuels supplied. Since the underlying assumptions regarding biomass feedstock availability, conversion technology costs, and future demand are highly uncertain, a significant portion of this dissertation explores the sensitivity of the model results to alternative assumptions.
The methodology has been developed to be flexible in the policy relevant questions that can be analyzed. However, due to the limited scope of this dissertation not the full breadth of research questions have not been
considered. The two case studies presented focus on the following questions for the United States and California in the next decade.
• What will the marginal cost be for producing and delivering biofuels as the quantity demanded changes (i.e., a biofuel supply curve)?
• How much biofuel can be produced relying only on waste and residue resources?
• Where will the industry take root?
• What is the cost and impact – air pollutant emissions and resource consumption – of waste biofuels in the Californian context?
Some of the additional research questions that should be taken into account as policies are developed to promote biofuels that can be asked using the model developed here are listed below.
• Where will the benefits and impacts occur?
• What will be the incremental capital cost of a transition to biofuels (i.e., capital investment required)?
• Are there tradeoffs between the biofuel production cost and environmental or societal benefits?
To be clear, I do not attempt to answer all of these questions in this dissertation.
1.3 Organization
The dissertation is organized as follows. Chapter 2 provides background information important to understanding the research questions. First, the stage is set for biofuel supply assessment by explaining the potential role of biofuels in the fuel sector, and how their production fits into the scheme of energy and agriculture. Previous work describing biofuel supply in both quantities and impacts is reviewed. Past methodological approaches for modeling future biofuel supplies are described. The strengths and
weaknesses of each approach are discussed, defining how the methodology developed here contributes to the field.
Chapter 3 lays out the methodology that has been developed. First, the framework of the model and the generic model formulation are given, followed by refinements on the formulation.
Chapter 4 describes the current status of biomass conversion technologies and describes the technology characterizations used in the subsequent case
studies. Chapter 5 reviews background literature and describes the data sets used for resource assessments, transportation cost models, and fuel demand.
In Chapters 6 and 7, case studies utilizing the methodology are described. Chapter 6 considers biofuel supply potential in the United States referenced to the year 2018 with a focus on meeting policy goals. Chapter 7 considers near term utilization of waste and residue biomass for biofuels production in California with an emissions accounting framework integrated with the spatial economic model. The two case studies highlight different challenges and benefits in utilizing the methodology. The national scale model uses county-level resource estimates with a spatial fuel demand constraints. The size of this analysis constitutes computational challenges resolved through the coupling of regional scale solutions. The California model demonstrates a high-resolution implementation of the approach with the explicit emissions accounting for the modeled biofuel industry.
Chapter 8 provides a summary of main findings, highlights the strengths, weaknesses and draws relevant conclusions regarding suitability of the modeling approach with recommendations for future enhancements.
2
BACKGROUND
2.1 Biofuel contextIn recent government policy in the United States and around the world, biofuels have been suggested as a cure for a number of ills caused by the transportation sector’s dependence on petroleum for energy (European
Parliament and Council, 2003; U.S. Congress, 2007). The most prominent of these are the reduction of greenhouse gas emissions in the face of global climate change (European Parliament and Council, 2003), enhancement of energy security (U.S. Congress, 2007) and supporting rural economies (European Parliament and Council, 2003; U.S. Congress, 2007). Secondary arguments have been made that the use of some biofuels would have lower air and water quality impacts compared to gasoline and diesel. In particular, ethanol has been used as an oxygenate for reformulated gasoline in order to reduce emissions of smog-forming compounds (Nadim et al., 2001).
From a broad-brush perspective, recent studies suggest that biofuels
appear capable of contributing to progress towards those policy goals over the next few decades. An economic model of United States agriculture found that domestic agricultural and forest resources could provide 60 billion gallons of ethanol and 1.6 billion gallons of biodiesel while also significantly increasing farm income and jobs in agriculture and renewable energy (De La Torre Ugarte et al., 2007). A study of the technical potential for “sustainable” cellulosic biomass production in the US was found to be 1.3 billion tons per
year, equivalent to approximately 30% of the US petroleum consumption by energy content (Perlack et al., 2005). A study of climate mitigation strategies from the agriculture and forestry sectors found that biofuels provide reductions of approximately 100 million metric tons of carbon
equivalent (MMTCE) at a carbon price of $100/MMTCE (McCarl et al., 2001). However, it is becoming increasingly clear that the attractiveness of
biofuels is dependent on the specific pathways used to produce them (Kim et al., 2005; Delucchi, 2006; Farrell et al., 2007; Turner et al., 2007; Unnasch et al., 2007; Zah et al., 2007). Even within corn ethanol production, there is a large range of potential direct greenhouse gas emissions and environmental impacts (Kim et al., 2005; Turner et al., 2007; Unnasch et al., 2007). Zah et al (2007) found a large range of both local environmental impacts and
greenhouse gas emissions when considering potential biofuel options for Switzerland. Many biofuel pathways demonstrated significantly worse
environmental performance than the petroleum fuels they would replace (Zah et al., 2007).
Consequently, there is a vigorous debate within the academic community and among government, environmental and industry groups regarding the sustainability of biofuel production – due to both environmental impacts and competition with food production. However, information that relates
sustainability to the supply potential is scarce. The definition of “sustainable biofuels” is neither clear nor agreed upon. Generally, the definition of
sustainable practice is one that meets current needs without
compromising the ability of future generations to meet their needs (WCED, 1987). But the generality of this definition leaves broad room for
interpretation in application to the questions surrounding biofuel production (Yeh et al., 2009).
There are many ways in which biofuels can be environmentally
unsustainable – habitat loss/deforestation, soil degradation, greenhouse gas emissions, pollution of water and air, aquifer depletion, etc. The production of energy crops and the conversion processes of all biofuels require significant water consumption, and many biofuel pathways can lead to reduction in water quality through intensification of agriculture (National Research
Council, 2008). The change in life cycle air pollutant emissions using biofuels compared to a baseline petroleum fuel depends on the particular biofuel pathway, with some yielding a net benefit and others a net detriment (Wu et al., 2005). There are also concerns about the soil quality impacts of
agricultural residue removal for use in biofuel production (Lal, 2005). And production of biofuels can pose a threat to biodiversity through habitat loss as well as water and soil quality impacts (Cook et al., 1991).
Competition for land between food and energy crops is also cause for caution. The boom in production of corn-based ethanol in response to both federal mandates and gasoline prices played a significant role in the doubling of the price of corn from 2006 to 2008 (Babcock, 2008). Most options to
produce biofuels on a significant scale will require the use of large
quantities of agricultural land. But productive agricultural land is a limited and valuable resource that provides the basic need of nourishment to a growing global population. The question of whether it is a good idea to incentivize the development of another major use for this scarce resource is becoming important, especially since many agricultural practices have negative environmental impacts.
Furthermore, introducing biofuel production that is competitive with petroleum fuels links the global agricultural and land markets to energy markets. It is not likely to be possible to limit production of biofuels to marginal land; biomass, like traditional crops, will grow better and be more profitable on good agricultural land. A potential danger in linking these markets is that it can give those with higher purchasing power the ability to meet their energy needs by indirectly starving those with lower purchasing power.
Although expanding the quantity of lands in agricultural production can ease the problem of direct food-fuel competition, this expansion often leads to major environmental impacts, including deforestation, habitat loss and resulting loss in biodiversity (Cook et al., 1991), as well as greenhouse gas emissions caused by releasing the carbon stocks of the converted land (Fargione et al., 2008; Searchinger et al., 2008). For many stakeholders in
biofuels policy, these impacts more than cancel the gains achieved by the production of biofuels.
Despite these serious issues, however, it is important to note that there is a great deal of variability in the potential impact of biofuel production
pathways – on both food production and the environment. Within this
variability, the opportunity exists for a limited sustainable biofuel industry. But the viability and extent of such a sustainable biofuels industry depends on the costs of production, primary and co-product market values, and any subsidies for such production influencing overall profits. The policy basis for the latter, in addition to mandates and other government influences,
therefore requires extensive information relating to net economic,
environmental, and social benefits, if any. The present debate over biofuels in part reflects high levels of uncertainty about these outcomes and the need for more comprehensive information.
The costs and impacts of producing biofuels depend on the geography of the resource to be exploited, the size of the biorefinery and the cost of
accessing the fuel market. These factors are not independent. For example, the economically optimal size of a biorefinery will depend on the spatial density of the resource it is exploiting, with dense biomass resources capable of supporting large biorefineries. As biomass resource supply becomes more dispersed, increasing feedstock transportation costs can outpace the scale economies of increasing biorefinery size. High costs to access fuel markets for
the sale of biofuel products can also make a low-cost producer less profitable than a producer with higher costs but nearer to the market. Geography is salient for the environmental impacts associated with biofuel production since the transportation of both biomass feedstock and product fuels can be significant for the life cycle impact of biofuel pathways (Wakeley et al., 2008).
A number of studies have considered the basic tradeoff in the design of biofuel supply chains between the size of a biorefinery – taking advantage of economies of scale – and the cost of biomass and biofuel transportation. Transportation of biomass is expensive relative to its value as an energy feedstock due to low energy density (Searcy et al., 2007). Thus in many cases, the additional transportation costs quickly outweigh opportunities for
economies of scale in the biorefinery, leading to a clearly defined optimal size of the biorefinery. However, the exact capacity of this optimal size is
situation dependent, with the spatial layout of the resource base, the scaling of the technology, purchasing agreements for feedstock (Kaylen et al., 2000) and the product market (Parker et al., 2008) all being relevant factors to consider. The spatial layouts of the resource and product markets in particular are not easily generalizable and vary considerably by locations.
In addition to the spatial aspects, competition for biomass may come from a number of sectors besides transportation fuels. Biomass production and conversion systems resulting in low lifecycle greenhouse gas emissions
(low-carbon) are considered attractive for a number of potential products in a carbon-constrained world. Electricity produced from biomass was found by Campbell et al. (2009) to be a more efficient use of biomass for the purpose of reducing carbon emissions than biofuels (Campbell et al., 2009). At present, it is unclear which of these products or combination of products will become the most attractive use of biomass. There are viable low or zero carbon alternatives in some sectors – such as wind and solar in the electricity sector – while other sectors that require energy dense liquid fuels – such as aviation and long haul freight – have fewer options and are likely to place the highest value on biomass as a feedstock.
The foregoing narrative illustrates that good policy will require an
improved understanding of biofuel systems, including the tradeoffs that exist between the size of the biofuel supply, economics and potential adverse
environmental and/or societal impacts. There has been little work done to show the quantity of biofuels that could be brought to bear on the
transportation energy system with clear accounting for cost estimation, technology choice, regional variations in supply, systems analysis of the full supply chain, environmental impacts and resource constraints, and the impact of potential regulations. I seek to fill this gap.
2.2 Approaches to modeling biofuel futures
A few different approaches have been taken to project future biofuel supplies. No method provides a satisfactory representation of all the
important aspects of the biofuel supply system, but each method can
leverage data and knowledge found using the other methods. Consequently, I classify the research to date into three categories. First, there are
assessments of biomass and/or biofuels using either technical estimates or economic models of the agriculture sector. Second are transportation fuel or energy sector economic models with limited description of resource supplies. Third are spatial infrastructure optimization models that find the optimal supply system for biomass-based fuels.
Technical estimates of biofuel potential have been performed at a number of scales using a range of limiting factors. Field et al (2008) developed a global estimate of biofuel potential using abandoned agricultural land that is not currently forested or urbanized. They found approximately 5% of the world primary energy could be provided by biofuels grown on these marginal lands. Other researchers using similar methods (Tillmann, 2006; Hoogwijk, 2003) found that a range of 2 - 35% of the energy demand could be met. Perlack et al (2005) estimated that 1.3 billion tons of biomass could become available in the United States by 2030 under optimistic scenarios of energy crop and agricultural residue production. Williams et al (2008) calculated that 32 of 83 million dry tons of the gross biomass produced in the state of California are technically available for energy production. These studies provide quantified resource assessments but do not account for the economics of biomass production and give little if any consideration to the conversion
technologies required to produce fuels. They are meant to provide rough estimates of the total sustainably available biomass resource only and do not provide any understanding of whether an economically viable industry is possible.
Economic models of the agriculture and forestry sectors improve upon technical assessments by capturing market effects. Two agricultural sector models have been developed with the purpose of answering questions about biomass as a potential energy and industrial feedstock in the United States and are describe in the following two paragraphs.
De la Torre Ugarte and Ray (2000) developed a dynamic, systems model of United States agriculture that is anchored to an externally provided baseline (such as FAPRI or USDA projections). A value for biomass as an energy feedstock along with estimated cost of production is introduced and the reaction of the agricultural market is simulated. The POLYSYS model has been used to project the impact on agricultural markets of producing 60 billion gallons per year of ethanol by the year 2030 (De La Torre Ugarte et al., 2007).
Khanna et al (2010) have developed a “dynamic multi-market equilibrium” model to consider the effects of policies on competition between food and fuel crops. The agricultural sector is modeled in detail with the introduction of switchgrass (Panicum virgatum) and miscanthus (Miscanthus giganteus) energy crops. Transport of the biomass and conversion to fuels are treated as
linear factors that convert biomass to fuel for a single set cost. The fuel market is simulated using elasticities of demand for gasoline and elasticity of substitution between gasoline and ethanol. Also, gasoline price is responsive to changes in demand as ethanol elbows its way into the market. Khanna and her co-authors make use of the detailed data within the model to report endogenously calculated emissions of greenhouse gases.
The Forest and Agricultural Sector Optimization Model – Greenhouse Gas Version (FASOMGHG) is an integrated economic model of the forest and agricultural sectors with a focus on land allocation decisions and subsequent impacts on greenhouse gas emissions (Daigneaultet al., 2009). It was used by the EPA in analyzing the RFS2 policy with two main goals (US EPA, 2010). First it provided the economic basis for the allocation of lands to energy crop production. Second it provided the domestic indirect land use change greenhouse gas emission component in analyzing the greenhouse gas impact of the fuel pathways (US EPA, 2010).
The economic models of the agricultural sector treat biomass as having a single value across all locations and types of biomass. This is not an accurate description of biomass for several reasons. Spatial markets will result from the high cost of biomass transportation and discrete locations of large
biomass consumers. Biomass producers located near the large consumers of biomass can demand higher prices for their biomass and therefore be more profitable than producers far from the consumers with the same costs of
production. Additionally, the term biomass refers to a heterogeneous set of materials of recent organic origin. This heterogeneity will be exploited to maximize the benefits of using biomass. The value of biomass generally varies with its end use. A bale of switchgrass has a value to a feedlot based on the nutritional content. The same bale is valued by its
cellulose/hemicellulose content by a biochemical ethanol producer and by its heating value (and ash properties) by thermochemical biofuel producers and electricity producers. These three aspects of a biomass feedstock are not proportional and different end users will value different biomass differently. In a simplified world, the end user with the highest value sets the price. Due to these two aspects of biomass it is important to consider both the value of the end use product and the location of the consumers of biomass when projecting supplies of biomass in a competitive market.
Other approaches to economic modeling of biomass and biofuels have focused on the energy market. Two studies to date demonstrate this
approach. Alfstad (2008) used the Department of Energy-Energy Technology Perspectives (DOE-ETP) MARKet ALlocation (MARKAL) model to analyze the likely outcome of the RFS2 biofuels mandate in the United States. MARKAL models are dynamic energy sector models with rich supporting information on technologies, resources and markets for energy products. The MARKAL framework uses a least-cost criterion for choosing between energy pathways to meet specified energy demands. In order to focus on biofuels,
Alfstad and his collaborators updated the biomass resource assessments and the technology models for biofuels production in the United States and countries most likely to export fuels to the United States and imposed the constraints of the RFS2 policy. The results predict that the mandate will not be met without overcoming significant barriers on the market and
infrastructure side of the equation. However, due to the nonspatial nature of these findings, they are more a reflection of assumptions in the model than analysis.
The BioTrans model has been developed to study biofuel transitions in Europe in reaction to policy mandates for biofuels (Lensink and Londo, 2010). It uses a least cost network flow modeling framework to choose biofuel
pathways in order to meet mandated production targets. Built with the purpose of analyzing biofuels, it makes several improvements while sacrificing complexity of market interactions in both the agricultural and energy markets. The spatial resolution is country-level for everything except the biomass supply, which is done at a sub national scale. It makes an
explicit characterization of marginal lands and the economics of potential energy crop production – yields, cost of production and revenue from an incumbent crop. Multiple technologies compete for resources and fuel market share. Each year is solved successively with installed capacity impacting the conversion cost through learning curves. Despite the focus of dynamics there is no consideration of sunk capital costs and existing
capacity. De Wit et al (2010) used BioTrans to demonstrate the potential for large market shares for biodiesel in the European market by 2030 and found that technological lock-in is a likely outcome if policies are not designed to diversify the market. The method is limited in that it uses set
transportation distances for intra-country deliveries and linear conversion costs.
To address the spatial aspects of biofuels production, a significant literature exists that focuses on understanding the best way to design bioenergy supply chains. A number of studies have focused on optimal biorefinery siting relative to the resource, given a standard biorefinery size (Graham et al., 2000; Zhan et al., 2005). Other studies explore the tradeoff between biorefinery size and feedstock transportation cost (Kaylen et al., 2000; Kumar et al., 2003).
Several recent papers have begun to address the design of a biomass-based industry in a full optimization framework. Freppaz developed a decision support system for the exploitation of forest resources considering multiple energy products and the spatial layout of both the supply and demand
(Freppaz et al., 2004). My previous work has included research on the siting and sizing of biomass hydrogen biorefineries exploiting California’s rice straw resource (Parker et al., 2008) and biofuels production in the western United Sates (Parker et al., 2010). Schmidt et al (2009) developed a spatially explicit supply chain optimization to compare the cost-effectiveness of CO2 emissions
reduction through heat, electricity or fuels production using woody
biomass in Austria. The methods used are similar to the methods presented in this dissertation. The main differences are that continuously variable biorefinery sizes are used here while Schmidt et al use discrete sizes and the objective is to minimize cost rather than maximize profit. Schmidt and his co-authors use a prescreening method to select potential biorefinery sizes. These studies have necessarily limited their scopes due to computational and data availability concerns. In order to be used for policy analysis these models need to have an expanded scope that studies large regions such as the United States or the European Union.
The biofuel infrastructure models borrow their analytical formulation from the field of facility location within the field of operations research. For a good background on the facility location problem, see Owen and Daskin (1998). Melo et al (2009) provides a recent review of supply chain management
studies with facility location. Melo points out that surprisingly few studies of facility location and supply chain management use a profit-maximizing
objective despite it being the presumed goal of all industries that are
modeled. The profit-maximizing objective described later is a key component to enable policy analysis within the framework.
2.2.1 Summary
The three approaches described above focus on different important traits of the biofuel pathway. The agricultural partial equilibrium models have the
strongest representation of the supply of energy crops incorporating the competition for scarce land resources between conventional food crops and energy crops. In general, they either ignore or simplify the infrastructure and conversion technology considerations. The infrastructure models provide the opposite, focusing on the important spatial features and layout of
biorefineries while using a simplified resource assessment and considering small regions. The “bottom-up” engineering-economic models bring in the dynamic aspect on the technology side and the full energy market but sacrifice the detailed spatial aspects of resource supply and biofuel system layout. The work presented in this dissertation is a spatially explicit
infrastructure model that has been demonstrated at the U.S. national scale. The main advantages of this approach are the following. First, explicit consideration is given to the tradeoff between economies of scale and
transportation costs that is constrained by real-world geographic information. This not only improves the estimate but also guarantees that the modeled system is anchored to a realistic supply system. Second, the use of a profit-maximizing framework allows greater flexibility in the types of questions that can be asked. For example, the impact of incentives in the form of subsidies can be analyzed or the impact of spatial variation in fuel prices can be considered. Finally, the data intensive approach based in engineering estimates of costs provides a relatively transparent and flexible model for
analyzing the sensitivity of the highly uncertain parameters involved in projecting future fuel supplies.
3
METHODOLOGY
The methodology used here is to build a series of scenarios varying policy, market and technology parameters that influence the design of the biofuel industry. The industry is then modeled using a spatially explicit integrated supply chain model. This model describes the optimal behavior of a biofuel industry given a fuel demand, biofuel selling price, and feedstock supply constraints. If biofuel can be delivered to the fuel terminals for less than the given selling price then it is profitable for the industry to supply that biofuel and the infrastructure is built to reap that profit. If biofuels cannot be delivered for less than the selling price then the fuel demand is met with conventional fuels at the given selling price. In addition, when demand for fuel exceeds the supply of feedstock, the difference is made up with
conventional fuels.
The model has been adapted to be responsive to policy and market conditions. This is made possible by (1) flexible spatially explicit resource and technology assessments, (2) a mixed integer-linear supply chain optimization model, (3) spatial models of transportation costs and (4) an environmental accounting model of emissions and resource consumption.
For each scenario, the optimal designs of the biofuel systems are found over a range of prices in order to produce supply curves. The supply curves show the quantity of fuel that would be made available at a given market price for biofuels. In economic terms they are considered long-run marginal
cost curves for biofuel production as they account for both capital and operating costs. Along with these optimized supply curves, estimates of
greenhouse gas emissions, emissions of criteria air pollutants, water demand, consumption of primary energy sources, land use changes, and types and quantities of biomass consumed can be made subject to data availability at each price point.
To develop optimal biofuel system designs, a number of models are integrated to work together, enabling a systemic view while maintaining computational feasibility. At the center of the integrated model is the supply chain optimization model that sites and sizes biorefineries, allocates the resources to the biorefineries and allocates the fuel produced to the demands. External models provide the input parameters for this optimization model. The resource is spatially characterized using a Geographic Information System (GIS) model that integrates and expands several existing resource assessments. Fuel demand is characterized using a spatial demand
assignment model and allocated to fuel distribution terminals.
Transportation cost calculations are performed in a GIS network model. The biorefinery cost and performance are described by a spreadsheet engineering model that simplifies the production costs into an integer-linear function of the fuel output and biomass inputs.
3.1.1 Major Strengths and Weaknesses of the Modeling Framework
The proposed approach is an engineering-centric method for developing supply curves. It focuses on the details of engineering costs, environmental impact accounting and spatial modeling. This approach enables meticulous analysis of the variation in environmental impacts, supply and cost of a variety of biofuel pathways with real-world geographies. Capturing the richness of this variation in biofuel pathways will provide insight in the degree to which biofuels can accomplish policy goals. The major weakness of this approach is that agricultural and energy markets are not endogenously considered.
The framework proposed here does not naturally lend itself to the study of economic feedback loops that the industry will create. First, demand for feedstock does not impact the modeled cost of acquiring the feedstock. Since the model maximizes total industry profit, a portion of the profit is expected to flow to the feedstock providers. Second, it is assumed that the modeled biofuels industry does not create any impact on the market price of the co-products. Because the co-product markets are not endogenously considered, sensitivity analysis is required. Finally, it is assumed that the consumption of biomass waste and residue streams does not impact the industries
producing the wastes and residue streams. Additionally, biorefineries are developed in a cooperative fashion, which maximizes profit for the industry
as a whole, not considering individual market actors. These limitations are due to simplifying assumptions for the model. The simplifications allow the model to focus on spatial aspects – resource and demand layout and infrastructure design – including the secondary market effects would require an iterative approach to finding market clearing prices that would lead to significantly longer computation times.
3.2 Overview
A geographically explicit biomass resource assessment and infrastructure network model is integrated with technoeconomic models of the conversion technologies and an emissions inventory model to provide analysis of
potential biofuel supply pathways. The analysis has five main components – 1) geographically-explicit biomass resource assessments, 2)
engineering/economic models of the conversion technologies, 3) models for multi-modal transportation of feedstock and fuels based on existing
transportation networks, 4) a supply chain optimization model that designs the fuel production system based on inputs from the other models, and 5) an emissions inventory model that calculates the emissions resulting from the designed supply chain. The optimization and emissions inventory models are described below.
Figure 1: Model organization and interaction of submodels 3.3 Supply chain model formulation
3.3.1 General formulation
The optimization model is formulated as a deterministic, multi-commodity, capacitated facility location problem. A biofuel supply chain optimization model was developed to consider explicit spatial distributions of biomass supply and fuel demands, competition among technologies for resources, and the economies of scale of conversion technologies in finding the best design for biofuel supply chains. The model locates, sizes, and allocates feedstock to biorefineries with the objective of maximizing the profitability of the industry as a whole. The profit considered is the sum of the profits for each individual feedstock supplier and fuel producer over the entire study region. Costs considered are those associated with feedstock procurement, transportation, conversion to fuel, and fuel transmission to distribution terminals. Fuel production and selling price determine industry revenue. The selling prices
of the product fuels are input parameters that are varied to create a supply curve.
Figure 2: Schematic of optimization model.
The model is formulated as a mixed integer linear program. Decisions integrated into the model are whether to build a biorefinery of a given
technology type, ‘t’, at a given site, ‘j’, (Xjt); if built, how many dry tons (US) of feedstock of type, ‘f’, is consumed per year by the biorefinery (Yfjft), the
quantity of fuel product type ‘p’ produced (Ybjpt) measured in millions of gallons per year (MGY), the fuel distribution terminals to which the fuel is delivered (Tjkt), and which feedstock supplies, located at location, ‘i’ at a particular procurement cost level, ‘c’, are exploited by the facility (Fijfc) measured in dry tons per year. Feedstock supply curves for each feedstock and supply point are defined at discrete cost levels in the model. These decisions are made for all potential sites simultaneously with no double
counting of resources. The objective of the program is to maximize the total annual profit of producing and delivering biofuels to distribution terminals. The profit is defined here as the annual revenue from the sale of biofuels and co-products less the annual cost of producing those biofuels. Table 2: Model variables and indices
Set Index Description Unit
i supply location
j potential biorefinery location k fuel terminal location
f feedstock type
t conversion technology p product type
c procurement cost level
e emission (CO, CO2, NOx, etc…) Variables
Fijfc Feedstock transported dry tons per year
Xjt Biorefinery built or not [0,1]
Yfjft Feedstock consumption dry tons per year
Ybjpt Product output gallons/kWh per year
Tjkp Product deliveries gallons per year
Table 3: Model parameters
Parameters Description Unit
Sifc Maximum available supply dry tons per year Dkp Maximum demand at terminal ‘k’ gallons per year
Pifc Procurement cost $/dry ton
TCijf Feedstock transport cost $/dry ton DCjkp Product transport cost $/gallon ajt Fixed biorefinery annualized cost $/year bjft Feedstock dependent biorefinery cost $/dry ton cjpt Product dependent biorefinery cost $/gallon or kWh MPkp Market price of product $/gallon or kWh
!fpt Conversion factor unit per dry ton
ggep Conversion factor for transforming all fuel products from volumetric units to energy units of gge
gge/gallon
Mjt Maximum biorefinery size dry tons per year ! Relaxation parameter for the proportional
blend requirement
"k Fraction of national vehicle miles traveled allocated to terminal ‘k’
#p Fraction of LDV fuel demand that can be met by fuel ‘p’
LDV fuel
demand Demand for light duty vehicle fuels in the analysis year Gallons of gasoline-equivalent per year
CP Carbon price $/ton CO2-eq
CIpt Carbon intensity of fuel product ‘p’ produced
using technology ‘t’ due to conversion process tons CO2-eq/gallon or kWh CIft Carbon intensity related to the production
and consumption of feedstock ‘f’ using technology ‘t’
tons CO2-eq/dry ton
FCifc Diesel consumed in harvest/ production of
biomass MMBtu/ton
EFem Emissions factor for emission ‘e’ per unit of
diesel fuel consumed by mode ‘m’ grams/MMBtu EFfe emissions factor for emission ‘e’ for feedstock
harvest/production from non-diesel inputs grams/ton feedstock EFte Emission factor for emission ‘e’ from the
conversion of biomass to fuel through technology ‘t’
grams/ton feedstock
FEm Fuel economy of transport by mode ‘m’ MMBtu/ton-mile dmij Miles by mode for each link in the feedstock
supply chain miles
$p Specific volume of product fuels gallons/ton
MCf Moisture content of feedstock type ‘f’ ton H2O/wet ton feedstock GWPe Global warming potential of emissions species tons CO2-eq/ton
The revenue is determined by the quantities and selling prices of the products (MPkp). In the model formulation, the energy products are
differentiated from other co-products. Non-energy co-products are included in the cost function (as negative variable cost) while energy co-products are part of the revenue. The costs considered are the procurement of feedstock (PCifc), the transportation of feedstock to the biorefinery (DCijf), the transportation of the product fuel to the distribution terminals (TCjkp) and the conversion cost. The conversion cost is dependent on the size of the biorefinery. I characterize it here as a binary-linear function with a fixed cost (at) if a facility is built and a variable cost (bt) dependent on the capacity of the biorefinery expressed in terms of feedstock input (Yfjft).
(1)
The objective function is combined with a number of constraints
representing the physical limitations or restrictions of the biomass industry in the mathematical model. The first set of constraints limit the biomass originating from a source at a price level to be less than the maximum supply of biomass of that type and price level at that source (Sifp) (equation 3).
(3)
The biofuel produced at a biorefinery is equal to the quantity of biofuel that can be produced from the biomass entering the biorefinery, given the
conversion efficiency (%ft) including handling loss (equation 4). I also
relate the biorefinery biomass input capacity to the biomass coming into the facility (equation 5) and the product fuels leaving the biorefinery to the production of biofuel at the biorefinery (equation 6).
(4)
(5)
(6) The size of the biorefinery must be zero if the fixed cost has not been paid (binary variable at that site is 0). If the binary variable is 1 then the
biorefinery can be no greater than its maximum allowable size for the technology (Mt) (equation 7).
(7) Fuel demand is limited at each terminal to represent either technical or policy constraints to the consumption of the fuel at that terminal. Different approaches for this spatial fuel demand constraint are discussed in section 3.3.2.
(8)
All variables representing physical quantities must take on either a
zero or positive value (equation 9). The binary variable for the existence of a biorefinery must take on a value of zero or one (equation 10).
(9)
(10) Each model run gives results of the industry-wide fuel production for a given price; which biorefinery locations are optimal and how big they are; and which biomass resources are used at each biorefinery. Multiple model runs are performed over a range of fuel prices. Plotting the industry production against fuel price gives the supply curve.
3.3.2 Approaches to spatial fuel demand constraint
Fuel demand at each terminal can be limited in a number of ways. The assumption used in the baseline model is that a proportion of fuel deliveries of a specific fuel type to each terminal must not be greater than ! more than the proportional vehicle fuel demand allocated to the terminal (equation 11). The parameter!provides the model a small degree of flexibility in fuel
deliveries. The choice of this parameter is a tradeoff between computational difficulty and the desired strictness of the constraint. The fuel demand is allocated by the fraction of the national VMT within the terminal’s service territory; this value is the parameter "#. Alternative formulations could include a blend wall (equation 12), where $p represents the allowable blend
fraction of a given fuel. Modeling E85 infrastructure requires changes to the objective to track the cost of installing E85 fuel pumps and additional constraints to track the quantities of ethanol used in E85 versus E10. I have not modeled in this analysis the required number of E85 stations in a region to accommodate full use of E85 in flex-fueled vehicles.
(11)
(12)
3.3.3 Approaches to handling greenhouse gas emissions
In a carbon-constrained world, an economic cost will exist for greenhouse gas emissions that must be accounted for in the profit equation of the biofuel industry. Two options have been explored for incorporating greenhouse gas emissions into the economic model. The first and simplest method uses default carbon intensity values for different classifications of fuels. For example, wet mill corn ethanol facilities can be given a value of 100 g MJ-1 of carbon dioxide equivalent emissions while ethanol from corn stover is given a value of 20 g MJ-1. Using this method converts the cost equation to Equation 13. This method requires established emission factors for each biofuel
pathway considered, which are not well known in some cases. Furthermore, existing emissions factors may not match the exact pathways being modeled and therefore are not true measures of the modeled biofuel supply.
(13)
Alternatively, emissions for the biofuel pathways modeled can be tracked within the model to provide accurate estimates of the specific pathways modeled down to the transportation distances and modes. This method presents its own difficulties in data requirements and ensuring consistency. It is also not used in the current regulatory environment. The California Air Resources Board, the US EPA and UK Renewable Transport Fuel Obligation all use a default and opt-in framework for determining the carbon intensity of a specific batch of fuel. The emissions tracking model is described by
Equation 14.
(14)
(15) The emission factor formulation (equation 13) is appropriate for analysis of the industry response to policies with default values for different pathways. It can also give a better representation of the carbon impacts in some cases. For example, the best value for the carbon intensity of a bushel of corn may not be the carbon intensity for the particular bushel used but rather the marginal bushel on the world market.
3.3.4 Alternative cost minimization model
The modeling approach taken here is to develop a profit-maximizing model. In many cases, a profit-maximizing model yields the same results as a cost minimization model. However, a profit maximizing model has several characteristics that make it more advantageous than cost minimization for this particular application.
The first advantage of profit maximization is the flexibility of constraints it allows. In cost minimization, some constraints must be predetermined that are not necessary for profit maximization. For example one must minimize cost subject to the full utilization of the resource or satisfying a
predetermined demand. These constraints are necessary to prevent the model from always producing a null answer. The constraints have the disadvantage of reducing the model’s flexibility. In choosing the optimal design, fractional levels of resource use and demand satisfaction may be the best option. This is especially true for modeling the biofuel industry which will account for a fraction of the fuel market into which the biofuels are sold. A profit maximizing approach avoids these issues by allowing the model to choose which resources to use and which demands to serve based on
balancing the costs of production of a good with the price of the good. The second advantage is in the interpretation of the results. A profit maximizing approach seeks to resolve the question about how much fuel can
be produced from a resource while recognizing the importance of market prices for answering the question.
A third advantage is that with mixed integer-linear models the marginal values are not reliably obtained. Economic theory tells us that the marginal cost is the interesting metric for evaluating cost of meeting a production target for any good. The profit maximizing model – as the dual problem to cost minimization – provides this information in a straightforward manner.
The last advantage of the profit maximizing method is that it allows for infrastructure design to respond to price differentials between demand centers. This feature can be used to replicate the disparate fuel prices currently seen across space or to evaluate regional policies that may attract biofuels to a region such as California’s Low Carbon Fuel Standard. The model can be used to evaluate the prices that California would need to pay in order to attract enough low carbon biofuels to meet the standard.
Despite the advantages of the profit-maximizing model, some research questions are better suited to a cost minimization approach of the model. Conversion from the profit-maximizing framework to a cost minimization is straightforward. The objective is replaced by an objective to minimize the cost and a binding constraint must be introduced.
Depending on the research question, either the supply or the demand can be binding constraints. In the case of a mandated volume of fuel, an
mandate? The total demand becomes the binding constraint, as seen in Equation 16. In some cases, the research may be interested in the least cost system for utilizing a certain resource. For example, the results of an
agricultural-economic model may give the production of biomass at $40/ton at the roadside. This assessment depends on all farmers being able to get that price and so all the resource must be used for a consistent biofuel supply assessment. In these cases, Equation 17 replaces the supply constraint (Equation 3) in the generic formulation.
(16)
(17) Greenhouse gas emissions may also provide a binding constraint for a cost minimization model. Either as an absolute reduction against the baseline gasoline or by expanding the system boundaries to include petroleum fuels production. As a reduction against the baseline the constraint can be formulated as Equation 18.
(18)
3.3.5 Competition with other biomass consuming industries
Competition for biomass feedstock between industries is an expected outcome of the combination of policy, market, and technology developments that seek to move away from fossil feedstocks for many sectors including electricity, fuels, plastics, and chemicals. These new and increased uses for
biomass will impact the cost of providing biofuels by providing a competing use for biomass. Each sector has its own characteristics of
quantities, technologies, yields and prices that influence the competitiveness of each for limited biomass resource. The generic model can easily
accommodate these uses of biomass, provided availability of the needed data. The nonfuel markets would be added as additional products from
biorefineries with the conversion technology models updated to include the technologies to produce the alternative biomass-based products. An example of this considering competition between electricity and biofuel sectors can be found in Tittmann, et al (2010).
3.3.6 Linking to results from agricultural economic models
Agricultural economic models are arguably the preferred method for a resource assessment for agricultural biomass. The predominant method for performing a resource assessment with agricultural economic models is to set a farm gate price for biomass as a perturbation of the existing agricultural system and find out how the introduction of this new commodity impacts the system. For the purposes of the proposed modeling framework this is
problematic. To remain consistent with results of agricultural economic models, only biomass corresponding to a single farm gate price can be used and all of the biomass available at a single farm gate price must be consumed if any of it is consumed. This type of resource assessment does not fit neatly into the modeling framework.
